SoundHound AI Is Betting $43 Million That a Dying Competitor Customers Will Save It
SoundHound AI just agreed to buy a company that lost $201 million over the past two years. Its stock is down 15 percent in 2026. And Wall Street is still telling investors to buy it.
The deal is SoundHound's acquisition of LivePerson, an all-stock transaction valuing LivePerson's equity at roughly $43 million — a 22 percent premium to its prior share price, but a valuation that reflects a business that has collapsed. LivePerson lost $134.2 million in 2024 and $67.2 million in 2025, according to The Motley Fool. Its stock has dropped nearly 100 percent over five years. The question SoundHound is asking investors to answer is not whether LivePerson is worth $43 million. It is whether a voice AI company that cannot yet make money can successfully absorb a voice AI company that is actively losing more of it.
The combined company would serve enterprise clients across more than 30 countries, including 12 of the top 15 global banks, four of the top five global airlines, and four of the top five global automakers, according to a Wedbush analysis. That customer footprint is the bull case in a single sentence: SoundHound is betting that the agentic AI era — the idea that software agents will increasingly handle customer service, sales, and technical support without human intervention — will drive enough incremental demand for voice AI infrastructure to make the LivePerson absorption profitable. Analysts surveyed by Zacks are projecting roughly 60 percent upside from current levels, with a consensus price target of $13.50 against a range of $8 to $20, per Zacks.
The case for SoundHound as infrastructure rests on enterprise integration work that general-purpose models have not replicated. GPT-4o and Gemini Live can hold a conversation. They do not natively handle the compliance logging, domain-specific training pipelines, and legacy system connectors that an enterprise deploying voice AI at scale requires. SoundHound's automotive and IoT segment — its original business — grew 88 percent year-over-year excluding acquisition impacts in the first quarter of 2026, per the earnings transcript, which suggests the integration work has real customers. That growth is also the bull-bear seam: if SoundHound's differentiation is integration specificity rather than model capability, the question is whether that specificity survives when general-purpose models add enterprise connectors over time.
The financial picture is complicated by a 19-point spread between reported margins. SoundHound's GAAP gross margin was 31 percent in the first quarter; its non-GAAP gross margin was 50 percent, according to the earnings transcript. That gap typically reflects stock-based compensation and other non-cash or one-time adjustments that companies exclude from alternative financial measures. Whether the GAAP figure represents structural margin compression or an accounting artifact matters: a 31 percent gross margin means every dollar of revenue retains 31 cents before operating expenses; a 50 percent non-GAAP figure suggests the underlying business is more profitable than the reported number. The transcript does not break down what drives the difference, and analysts covering the stock have not published a detailed GAAP-to-non-GAAP reconciliation in the sources available.
SoundHound is not a profitable business trying to get larger. It lost $26.7 million on adjusted EBITDA in the first quarter of 2026, on revenue of $44.2 million — revenue that was itself up 52 percent year-over-year. The company has not disclosed whether any LivePerson enterprise customers have publicly committed to remain with the combined company through close. No major bank, airline, or automaker has announced plans to exit, according to the available record. The retention question — whether the logos on the combined company's client list represent revenue that stays or revenue that walks — is the load-bearing fact for the bull thesis. The deal structure requires SoundHound to retain those customers while absorbing LivePerson's accumulated losses.
The 2026 revenue guidance is $225 million to $260 million. Post-acquisition 2027 targets reach $350 million to $400 million, with at least $100 million of that attributed to LivePerson's global customer base, per the earnings transcript. In other words, the $43 million price tag comes with a $100 million revenue target attached to it. Whether that target is achievable depends on whether LivePerson's enterprise clients stay and whether they expand their spending with the combined company.
What happens next is a test of customer retention. If the banks, airlines, and automakers that make up LivePerson's legacy base remain through close and begin purchasing additional SoundHound services, the agentic AI thesis survives contact with the income statement. If they do not, SoundHound will have paid $43 million in stock for the right to absorb $201 million in losses and whatever revenue attrition follows. The stock is down 15 percent in 2026 because the market is not yet convinced the first outcome is more likely than the second.